Presentation on theme: "NC Energy Estimators Philip Rodrigues. Issues Need to choose what true E to estimate. Options are: –showerEnergy –trueVisibleE –y*E_nu Need to choose."— Presentation transcript:
Issues Need to choose what true E to estimate. Options are: –showerEnergy –trueVisibleE –y*E_nu Need to choose what reco variable to use. Options galore. Once chosen, calibrate it. –For true NC only, or for background too?
Method Hedge my bets: –Run with different E_true –Run with/without CC bg –Run with different E_reco 1.Plot (E_reco-E_true)/E_true 2.Fit gaussian to peak 3.Scale E_reco until fit peaks at 0 All ND MC so far, until fixed FD MC becomes available
Distributions - Uncalibrated E_reco=event.energyGeV. NC selected by Tom’s cuts Fiducial, high mult clean Fitting peak biases towards calibrating NC True CC events selected as NC tend to have energy overestimated
Distributions – calibrated for NC Fit gaussian to NC selected, true NC True NC well fit, slight bias in background
Choice of E_reco RMS after calibration as measure of width EstimatorTrue NC only+True CC event.energyGeV0.4910.579 reco.nuEnergy= reco.nuEnergyCC 0.4660.502 reco.showerEnergyCC0.5180.504 reco.showerEnergyNC0.3870.463 shower.deweightCCGeV0.3660.401 shower.deweightNCGeV = reco.nuEnergyNC 0.3870.463 shower.linearCCGeV0.5170.508 shower.linearNCGeV0.4850.504
True NC vs all selected: How big is the difference? Phill uses E_true=E_shw + E_mu Scale factors for different choices: deweightNCGeVlinearNCGeV -True CC, -E_mu0.9010.928 +True CC, -E_mu0.9050.902 -True CC, +E_mu0.9080.905 +True CC, +E_mu0.9100.909